Fix splitting non-ASCII strings if unicode_strings feature is enabled

This commit is contained in:
Petr Písař 2017-08-09 14:50:40 +02:00
parent 2d9a38d99f
commit a0ea65ae26
2 changed files with 233 additions and 0 deletions

View File

@ -0,0 +1,226 @@
From 5aca16e032861ea3dfcc96ad417ea87e2b1552e5 Mon Sep 17 00:00:00 2001
From: Aaron Crane <arc@cpan.org>
Date: Sat, 4 Mar 2017 12:50:58 +0000
Subject: [PATCH] RT #130907: Fix the Unicode Bug in split " "
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Ported to 5.26.0:
commit 20ae58f7a9bbf84d043d6e90f5988b6e3ca4ee3d
Author: Aaron Crane <arc@cpan.org>
Date: Sat Mar 4 12:50:58 2017 +0000
RT #130907: Fix the Unicode Bug in split " "
Signed-off-by: Petr Písař <ppisar@redhat.com>
---
lib/feature.pm | 5 +++--
pod/perldelta.pod | 9 +++++++++
pod/perlfunc.pod | 8 ++++++++
pod/perlunicode.pod | 11 +++++++++++
pod/perluniintro.pod | 5 +++--
pp.c | 13 +++++++++++++
regen/feature.pl | 5 +++--
t/op/split.t | 20 +++++++++++++++++++-
8 files changed, 69 insertions(+), 7 deletions(-)
diff --git a/lib/feature.pm b/lib/feature.pm
index ed13273..93e020b 100644
--- a/lib/feature.pm
+++ b/lib/feature.pm
@@ -175,8 +175,9 @@ C<use feature 'unicode_strings'> subpragma is B<strongly> recommended.
This feature is available starting with Perl 5.12; was almost fully
implemented in Perl 5.14; and extended in Perl 5.16 to cover C<quotemeta>;
-and extended further in Perl 5.26 to cover L<the range
-operator|perlop/Range Operators>.
+was extended further in Perl 5.26 to cover L<the range
+operator|perlop/Range Operators>; and was extended again in Perl 5.28 to
+cover L<special-cased whitespace splitting|perlfunc/split>.
=head2 The 'unicode_eval' and 'evalbytes' features
diff --git a/pod/perldelta.pod b/pod/perldelta.pod
index 06dcd1d..d31335f 100644
--- a/pod/perldelta.pod
+++ b/pod/perldelta.pod
@@ -3206,6 +3206,15 @@ calls.
Parsing bad POSIX charclasses no longer leaks memory.
L<[perl #128313]|https://rt.perl.org/Public/Bug/Display.html?id=128313>
+=item *
+
+C<split ' '> now correctly handles the argument being split when in the
+scope of the L<< C<unicode_strings>|feature/"The 'unicode_strings' feature"
+>> feature. Previously, when a string using the single-byte internal
+representation contained characters that are whitespace by Unicode rules but
+not by ASCII rules, it treated those characters as part of fields rather
+than as field separators. [perl #130907]
+
=back
=head1 Known Problems
diff --git a/pod/perlfunc.pod b/pod/perlfunc.pod
index b8dca6e..9abadf4 100644
--- a/pod/perlfunc.pod
+++ b/pod/perlfunc.pod
@@ -7616,6 +7616,14 @@ special case was restricted to the use of a plain S<C<" ">> as the
pattern argument to split; in Perl 5.18.0 and later this special case is
triggered by any expression which evaluates to the simple string S<C<" ">>.
+As of Perl 5.28, this special-cased whitespace splitting works as expected in
+the scope of L<< S<C<"use feature 'unicode_strings">>|feature/The
+'unicode_strings' feature >>. In previous versions, and outside the scope of
+that feature, it exhibits L<perlunicode/The "Unicode Bug">: characters that are
+whitespace according to Unicode rules but not according to ASCII rules can be
+treated as part of fields rather than as field separators, depending on the
+string's internal encoding.
+
If omitted, PATTERN defaults to a single space, S<C<" ">>, triggering
the previously described I<awk> emulation.
diff --git a/pod/perlunicode.pod b/pod/perlunicode.pod
index 9c13c35..2e84e95 100644
--- a/pod/perlunicode.pod
+++ b/pod/perlunicode.pod
@@ -1835,6 +1835,17 @@ outside its scope, it could produce strings whose length in characters
exceeded that of the right-hand side, where the right-hand side took up more
bytes than the correct range endpoint.
+=item *
+
+In L<< C<split>'s special-case whitespace splitting|perlfunc/split >>.
+
+Starting in Perl 5.28.0, the C<split> function with a pattern specified as
+a string containing a single space handles whitespace characters consistently
+within the scope of of C<unicode_strings>. Prior to that, or outside its scope,
+characters that are whitespace according to Unicode rules but not according to
+ASCII rules were treated as field contents rather than field separators when
+they appear in byte-encoded strings.
+
=back
You can see from the above that the effect of C<unicode_strings>
diff --git a/pod/perluniintro.pod b/pod/perluniintro.pod
index d35de34..595ec46 100644
--- a/pod/perluniintro.pod
+++ b/pod/perluniintro.pod
@@ -151,11 +151,12 @@ serious Unicode work. The maintenance release 5.6.1 fixed many of the
problems of the initial Unicode implementation, but for example
regular expressions still do not work with Unicode in 5.6.1.
Perl v5.14.0 is the first release where Unicode support is
-(almost) seamlessly integrable without some gotchas. (There are two
+(almost) seamlessly integrable without some gotchas. (There are a few
exceptions. Firstly, some differences in L<quotemeta|perlfunc/quotemeta>
were fixed starting in Perl 5.16.0. Secondly, some differences in
L<the range operator|perlop/Range Operators> were fixed starting in
-Perl 5.26.0.)
+Perl 5.26.0. Thirdly, some differences in L<split|perlfunc/split> were fixed
+started in Perl 5.28.0.)
To enable this
seamless support, you should C<use feature 'unicode_strings'> (which is
diff --git a/pp.c b/pp.c
index cc4cb59..d9dd005 100644
--- a/pp.c
+++ b/pp.c
@@ -5740,6 +5740,7 @@ PP(pp_split)
STRLEN len;
const char *s = SvPV_const(sv, len);
const bool do_utf8 = DO_UTF8(sv);
+ const bool in_uni_8_bit = IN_UNI_8_BIT;
const char *strend = s + len;
PMOP *pm = cPMOPx(PL_op);
REGEXP *rx;
@@ -5826,6 +5827,10 @@ PP(pp_split)
while (s < strend && isSPACE_LC(*s))
s++;
}
+ else if (in_uni_8_bit) {
+ while (s < strend && isSPACE_L1(*s))
+ s++;
+ }
else {
while (s < strend && isSPACE(*s))
s++;
@@ -5857,6 +5862,10 @@ PP(pp_split)
{
while (m < strend && !isSPACE_LC(*m))
++m;
+ }
+ else if (in_uni_8_bit) {
+ while (m < strend && !isSPACE_L1(*m))
+ ++m;
} else {
while (m < strend && !isSPACE(*m))
++m;
@@ -5891,6 +5900,10 @@ PP(pp_split)
{
while (s < strend && isSPACE_LC(*s))
++s;
+ }
+ else if (in_uni_8_bit) {
+ while (s < strend && isSPACE_L1(*s))
+ ++s;
} else {
while (s < strend && isSPACE(*s))
++s;
diff --git a/regen/feature.pl b/regen/feature.pl
index 579120e..8a4ce63 100755
--- a/regen/feature.pl
+++ b/regen/feature.pl
@@ -485,8 +485,9 @@ C<use feature 'unicode_strings'> subpragma is B<strongly> recommended.
This feature is available starting with Perl 5.12; was almost fully
implemented in Perl 5.14; and extended in Perl 5.16 to cover C<quotemeta>;
-and extended further in Perl 5.26 to cover L<the range
-operator|perlop/Range Operators>.
+was extended further in Perl 5.26 to cover L<the range
+operator|perlop/Range Operators>; and was extended again in Perl 5.28 to
+cover L<special-cased whitespace splitting|perlfunc/split>.
=head2 The 'unicode_eval' and 'evalbytes' features
diff --git a/t/op/split.t b/t/op/split.t
index d60bcaf..038c5d7 100644
--- a/t/op/split.t
+++ b/t/op/split.t
@@ -7,7 +7,7 @@ BEGIN {
set_up_inc('../lib');
}
-plan tests => 163;
+plan tests => 172;
$FS = ':';
@@ -480,6 +480,24 @@ is($cnt, scalar(@ary));
qq{split(\$cond ? qr/ / : " ", "$exp") behaves as expected over repeated similar patterns};
}
+SKIP: {
+ # RT #130907: unicode_strings feature doesn't work with split ' '
+
+ my ($sp) = grep /\s/u, map chr, reverse 128 .. 255 # prefer \xA0 over \x85
+ or skip 'no unicode whitespace found in high-8-bit range', 9;
+
+ for (["$sp$sp. /", "leading unicode whitespace"],
+ [".$sp$sp/", "unicode whitespace separator"],
+ [". /$sp$sp", "trailing unicode whitespace"]) {
+ my ($str, $desc) = @$_;
+ use feature "unicode_strings";
+ my @got = split " ", $str;
+ is @got, 2, "whitespace split: $desc: field count";
+ is $got[0], '.', "whitespace split: $desc: field 0";
+ is $got[1], '/', "whitespace split: $desc: field 1";
+ }
+}
+
{
# 'RT #116086: split "\x20" does not work as documented';
my @results;
--
2.9.4

View File

@ -212,6 +212,10 @@ Patch52: perl-5.26.0-perl-131588-be-a-little-more-careful-in-arybase-_tie
# in upstream adter 5.27.1
Patch53: perl-5.27.1-perl-131597-ensure-the-GV-slot-is-filled-for-our-foo.patch
# Fix splitting non-ASCII strings if unicode_strings feature is enabled,
# RT#130907 in upstream after 5.27.1
Patch54: perl-5.27.1-RT-130907-Fix-the-Unicode-Bug-in-split.patch
# Link XS modules to libperl.so with EU::CBuilder on Linux, bug #960048
Patch200: perl-5.16.3-Link-XS-modules-to-libperl.so-with-EU-CBuilder-on-Li.patch
@ -2795,6 +2799,7 @@ Perl extension for Version Objects
%patch51 -p1
%patch52 -p1
%patch53 -p1
%patch54 -p1
%patch200 -p1
%patch201 -p1
@ -2833,6 +2838,7 @@ perl -x patchlevel.h \
'Fedora Patch51: Fix error message for "our sub foo::bar" (RT#131679)' \
'Fedora Patch52: Fix executing arybase::_tie_it() in Safe compartement (RT#131588)' \
'Fedora Patch53: Fix handling attribute specification on our variables (RT#131597)' \
'Fedora Patch54: Fix splitting non-ASCII strings if unicode_strings feature is enabled (RT#130907)' \
'Fedora Patch200: Link XS modules to libperl.so with EU::CBuilder on Linux' \
'Fedora Patch201: Link XS modules to libperl.so with EU::MM on Linux' \
%{nil}
@ -5127,6 +5133,7 @@ popd
- Fix error message for "our sub foo::bar" (RT#131679)
- Fix executing arybase::_tie_it() in Safe compartement (RT#131588)
- Fix handling attribute specification on our variables (RT#131597)
- Fix splitting non-ASCII strings if unicode_strings feature is enabled (RT#130907)
* Sat Jul 29 2017 Igor Gnatenko <ignatenkobrain@fedoraproject.org> - 4:5.26.0-397
- Enable separate debuginfo back